Final Review

• The objective of the class
• Philosophy, history, applications
• Mathematics
• Experiments, practical experience with tools
• The topics covered included

  The concept of agent, problem formulation

• agent.htm
• Good for short answers: One or two sentences
• Formulate the problem
• Decide evalauation mechanism
• Decide characteristics of an agent
• Decide the environment
• Questions such as "What is dynamic/accessible/episodic environment?"

  Searching, genetic algorithms

• search.htm and search1.htm
• Different types of searches
• Possibly descriptive question
• Given a particular tree show how certain search will proceed
• Time requirements are important
• Search problems generally deal with exponential time. You should be aware of time requirements

  Logic

  Pure logic

• propositional logic, first order logic, Horn clause logic, Higher order logic
• Translate English language to first order logic

  Fuzzy logic

• Brief questions
• Given pictures show how fuzzy membership is determined and used in reasoning

  Expert systems

• expert.htm
• Brief questions
• Ask for an example of how reasoning works in Expert system

  Probabilistic logic

• Bayes rule: The complicated form.
• Why is it difficult to apply Bayes rule without making certain independence assumptions
• Why do we need Belief nets
• Given a tree, show corresponding formula
• Given a formula, show corresponding tree (see the quiz). Chapter 15: first few pages
• What is the advantage of the belief network? Local computing: good for parallel processing

  Belief functions

• How they are different from probability?
• Why they may be more meaningful?
• Given a particular situation, construct a belief function.
• A thermometer tells you the temperature to be 0C. What is your belief that the roads outside are slippery? Reliability of the thermometer is 95%

  Bel({not Slip})=0.95, Bel({Slip}) = 0

  Bel({Slip,not slip} = 1

  Learning

• Learning decision making strategies from existing historical database

  Decision trees

• bring your calculators, you may be asked to calcualte the entropies (information)
• Descriptive questions

Rough sets

• Descriptive questions

• Other descriptive questions on misc. information mentioned in learn.htm

Neural networks

• All the key terminology for brief answers
• Some one paragraph answers
• Given a network file, you should be able to draw the network
• Communication
• Descriptive questions
• Given EBNF translate to FOL
• It is expected that you use your own evaluation